In a world increasingly shaped by artificial intelligence, few companies have left a mark in 2024 like the open-source project Hugging Face.

What began as a chatbot app has since evolved into a hub for open-source AI, becoming an indispensable resource for researchers, developers, and businesses alike. By 2023, following several investment rounds, Hugging Face was valued at $4.5 billion.

Hugging Face is Emerge’s Project of the Year 2024 for its transformative role in AI and dedication to democratizing machine learning. With visionary leadership, open-source tools, and a strong focus on ethics, it empowers researchers and startups worldwide. Thanks also to a thriving online community of open-source AI enthusiasts, Hugging Face has become a standard-bearer for responsible and collaborative AI innovation.

What is Hugging Face?

Hugging Face, founded in 2016 by French entrepreneurs Clément Delangue, Julien Chaumond, and Thomas Wolf and based in New York City, is an open-source platform for machine learning and natural language processing.

Consisting of a massive library of over one million AI models, 190,000 datasets, and 55,000 demo apps, Hugging Face lets developers, researchers, and data scientists build, train, share, and deploy AI models.

“We started as a gaming company, and discovered we could have a much larger impact when starting to open-source some of our research code. That led to our transformers library and seeing the impact and excitement about it in the community,” co-founder and Chief Science Officer Wolf told Decrypt. “We think open-source is the key approach to democratize machine learning.”

At its core is the transformers library, which offers state-of-the-art pre-trained models for a wide range of tasks. Users can explore models through browser-based inference widgets, access them via API, and deploy them across computing environments. Hugging Face also fosters collaboration by allowing users to share and fine-tune models through its Hub, a central repository where users can experiment with and contribute to cutting-edge AI models.

Fine-tuning in AI refers to taking a pre-trained AI model—which contains weights and features learned from initial datasets to train the model—and adapting it to perform a specific task, or improve performance on a specialized dataset.

“Open science and open-source AI prevent blackbox systems, make companies more accountable, and help [solve] today’s challenges—like mitigating biases, reducing misinformation, promoting copyright, and rewarding all stakeholders including artists and content creators in the value creation process,” co-founder and CEO Delangue said on X (formerly Twitter).

Democratizing AI

A common refrain in the decentralized and open-source community is “democratizing AI,” or empowering individuals to use AI for social good, innovation, and solving complex problems without the control of corporations and governments.

In an industry dominated by proprietary technologies and closed ecosystems, Hugging Face stands out for making cutting-edge tools freely available to the global AI community. Delangue reiterated Hugging Face’s commitment to the cause of democratizing AI during a June 2023 congressional hearing of the Committee on Science, Space, and Technology.

“Hugging Face is a community-oriented company based in the U.S. with the mission to democratize good machine learning,” Delangue said during the hearing. “We conduct our mission primarily through open source and open science, with our platform for hosting machine learning models and datasets, and an infrastructure that supports research and resources to lower the barrier for all backgrounds to contribute to AI.”

Democratizing AI is particularly impactful in underrepresented regions and industries, where researchers and small startups often lack the resources to compete with tech giants.

“The long-standing and widening resource divides, especially between industry and academia, limit who is able to contribute to innovative research and applications,” Delangue told Congress. “We strongly support the U.S. National AI Research Resource and resourcing small businesses and startups conducting public interest research.”

Collaboration over competition

Emphasizing Hugging Face’s collaborative spirit, the company has worked with other big names in AI, including Google, AWS, Meta, Nvidia, and Microsoft.

In January, Hugging Face teamed up with Google Cloud by combining its own open models with Google’s infrastructure, all with the goal of making AI more accessible. That same month, Hugging Face introduced its Hallucinations Leaderboard, which the company launched to address the ongoing problem of AI hallucinations.

“The challenge now is to have enough startups and teams ready to deploy models in various verticals,” Wolf said. “No need to wait for GPT-5; it’s time to build AI applications now by learning how to use, evaluate, and adapt these models in today’s world.”

In May, Hugging Face expanded its partnership with Microsoft that began back in 2022, providing developers with broader infrastructure and tools to create more powerful versions of their Copilot AI models. Later that month, Amazon announced a new alliance with Hugging Face to make it easier for developers to run AI models using Amazon’s computer chips.

Computer chip giant Nvidia announced a collaboration with Hugging Face in July that would bring its Nvidia-accelerated inference services to the open-source platform, enabling developers to deploy AI models like Llama 3 with up to five times faster token processing.

In October, Hugging Face launched HuggingChat, the platform’s answer to OpenAI’s ChatGPT. HuggingChat lets users choose among a diverse pool of open-source AI models for its text generation capabilities. That was followed by the release of Hugging Face Generative AI Services, or HUGS, which lets developers deploy and train AI models offline in a personalized environment.

At the Conference for Robot Learning in Germany in November, Hugging Face and NVIDIA announced a partnership to push open-source robotics forward, by combining Hugging Face’s robotics platform LeRobot with NVIDIA’s AI tools to blend simulation and real-world training—all with the goal of making robots smarter and more effective.

It hasn’t always been smooth sailing for Hugging Face, however. In November, the company faced backlash after it was revealed that a dataset with over a million posts was created using scraped content from the rising Bluesky social media platform before being removed the next day.

“I’ve removed the Bluesky data from the repo. While I wanted to support tool development for the platform, I recognize this approach violated principles of transparency and consent in data collection,” Hugging Face Machine Learning Librarian Daniel van Strein wrote on Bluesky. “I apologize for this mistake.”

The future of Hugging Face

Moving into 2025, Hugging Face’s CEO laid out his predictions for the coming year in AI—including the first major public protest related to AI, a major company’s market capitalization getting cut in half due to AI, and over 100,000 personal AI robots going up for pre-order.

“We will begin to see the economic and employment growth potential of AI, with 15 million AI builders on Hugging Face,” Delangue tweeted.

Wolf shared a similarly optimistic view of the future of open-source AI and robotics moving into 2025, pointing to more energy-efficient models, open-

“Many things excite me about the future but to name only a few,” Wolf said. “Smaller models that can be much more energy efficient, the rise of open-source robotics and the extension of all the tools we’ve discovered in AI to the field of science, for example, weather prediction, and material discovery.”

Hugging Face played a pivotal role in AI’s evolution in 2024 by driving innovation, global accessibility, and transparency while lowering barriers for startups and developers to create a multitude of AI solutions.